How to choose the right tape library

Choosing an expandable, modular tape library with sophisticated management features to support your enterprise can be difficult. Here's how to make the right choice.

Tape libraries are ideal for long-term storage of archives or backups. They are also a major capital expense and there are many things you need to consider when purchasing one. In this context, tape libraries are a step up, sometimes a long step up, from autoloaders. Autoloaders are simpler, less-expensive devices with a single robotic arm to move tapes, usually one tape drive at a time. Libraries cost more -- sometimes a lot more -- but they have higher capacities, more features and are generally much more reliable. For a business or a branch office that doesn't need much capacity, an autoloader is an appropriate step up from a single tape drive directly connected to a server. But trying to use an autoloader to do a tape library's job is false economy.

The hardware differences between brands of tape libraries have shrunk as the tape library market has grown. Hardware is still important, especially in areas like speed and bandwidth, but that's becoming more of what model you choose rather than what vendor. While some vendors, such as Sun Microsystems Inc., concentrate on the high end of the market, and others aim for the small- to medium-sized business (SMB) segment, nearly every vendor offers a range of libraries with speeds and capacities that cover most of the market.

Tape backup for remote offices tutorial

Part 1: Tape still has a a place in remote backup

Part 2: How to get the most out of your data backup tapes: Caring for magnetic media

Part 3: How to optimize your backup tape rotation strategy

Part 4: How to choose the right tape library

"Ten years ago, this [the tape library] was a very specialized part of the market," says Peri Grover, director of product management for tape automation at Overland Storage. "It's become more and more commoditized. The increasing value is in how easy it is to use [the library's] services, the integration and support."

Unlike the hardware, tape library management and integration are not nearly as  standardized and different vendors offer a rich selection of features. A good part of choosing a tape library is deciding what features are important to your system. In fact, it is likely to be a key differentiator in your choice of a tape library.

Compatibility with your systems

The most basic factor in choosing a tape library is compatibility with the rest of your system, especially your backup software. All the features in the world don't matter if you can't get the tape library to work with your system. Tape library vendors provide extensive compatibility lists of which backup software and hardware will work with their systems. Make sure the manufacturer of the library and the software guarantee compatibility with your system. But vendors are sometimes slow to update their compatibility lists and it pays to check with the vendor if the library isn't on their list. Make sure all the features that are important to you will work with your software/library combination.

The management and integration of a tape library

Management is an area where vendors have taken major strides in the last few years. A modern tape library is a lot smarter, a lot easier to manage and, generally, a lot easier to use than the products of the 1990s. "It's remarkable how much libraries have changed in doing intelligent things like grouping media and the ability to partition library to share across environments," says Molly Rector, vice president of marketing and product management at Spectra Logic Corp.

Part of this transformation is the changing nature of what's stored on tape. Rather than just dumping an image to tape, companies today tend to try to manage their data backup and archiving to make things faster to find by setting up separate sub-libraries for different applications or classes of data. In addition, mixed environments with Windows and Linux machines are more common as tape systems are being consolidated, leading to the need to store data from several kinds of systems in the same tape library.

For example, remote operation capability has become increasingly common in tape libraries, including the ability to manage and configure the library remotely and even eject tapes for offsite storage with a remote command.

However, these ease-of-use features aren't always found where the user needs them. One of the ironies of tape libraries is that the high-end products usually have much better ease-of-use and usability features than the low end ones. Considering that the people who use low-end libraries are likely to need the most help, this is particularly unfortunate. The good news is that, again, features vary widely among libraries and you can often find the features you want in your price range.

Tape encryption

If you ever take a tape offsite, it should be encrypted. Today, almost all tape library vendors offer some encryption options. You can encrypt on the client device, on the backup server, in the network, in a special appliance or in the library or on the tape itself. Some tape technologies, like LTO-4, come with built-in encryption features. Backup software from vendors like Symantec Corp. has encryption features. Vendors like Decru (a NetApp company) offer standalone encryption appliances.

However, just because encryption is now a standard best practice, this doesn't necessarily mean it should be done in the tape library. Spectra Logic, along with a number of other vendors, offers encryption in the tape library.

Others, like Overland, suggest encrypting the data before it arrives at the tape library. "The right place to do encryption is in the ISV [backup] application," Overland's Grover says. "That's the one place where everything is pretty standard."

Many users have tried and abandoned performing encryption with the backup application because of the hit on backup performance. But, according to Grover, the big issue is not performance. Modern encryption techniques, when properly applied, can be performed with little or no penalty in backup speed. Instead, the problem is ensuring you can recover the encrypted data when you need it.

"It sounds great at first, but you need to consider what it means," says Grover. "The ability to encrypt is one thing, but you really need a robust key management solution. Particularly if you have more than one site, do you have solution that scales across my multiple sites? People like encrypted data, but if a key gets lost there's no way to unencrypt that data. You need robust, standardized key management."

Key management still isn't standardized and standards for storage key management are only beginning to emerge. However, key management is increasingly integrated into tape library software as well as backup products such as Hewlett-Packard (HP) Co. StorageWorks. In other words, key management is widely available -- it just isn't standardized. Assuming the site or sites where you will be recovering have appropriate access to the keys, and compatible hardware and software, there shouldn't be a problem.

Tape library expandability

Expandability is one of the few real hardware differentiators left with tape libraries. Libraries differ strongly in how expandable they are, how many tapes and drives they will ultimately support, and the size of each expansion module.

Expandability is an important feature for handling data growth. By adding modules as needed, you can spread out the expense of the library rather than having to purchase something that is oversized for your needs. At worst, expandability becomes a get-out-of-jail-free (or cheap) card if you have seriously underestimated your needs.

Speed and capacity

In dealing with tape, "speed" refers to two things. On one hand, it means the speed at which the library can write data on tape. On the other, it refers to how quickly the balance of the system, especially the network, can feed data to the library.

It doesn't do any good to purchase a library that is faster than the network that is feeding it. Of course, this is determined not just by the speed of the tape drives but by the overall speed of the library, including the cycle times when the library needs to change a tape, so it will be slower than the theoretical speed of the tapes. Still, many libraries can absorb data faster than many networks can transmit it. Unless you plan on upgrading your network, this represents an unnecessary expense.

A tape library also needs to be appropriately sized to handle the amount of data you will be backing up. In theory, sizing a tape library is easy. You figure out how large your backup set will be, how many sets you will want to keep and add something for data growth. The practice is likely to be a whole lot messier. For starters, your backup set is almost certainly not the same size as the total amount of storage on your system. There's undoubtedly a lot of stuff that doesn't need to be backed up and if you use something like data deduplication, your backups may be a lot smaller.

Compression is also difficult to figure exactly. Vendors typically talk in terms of 2:1 compression, but that's an average, and a rather optimistic one at that. Unless you're storing a lot of highly redundant data, such as uncompressed pictures, 2:1 is the absolute best you're going to do and you probably won't do that well.

Encryption can also distort compression. If the data is properly encrypted, it has no redundancy so it can't be encrypted. Once the data has been compressed it can be encrypted nicely, but the rule is: Encryption first, compression second. That means if you're going to do encryption before the data arrives at the tape library, you can't use the library's built-in compression.

The backup architecture enters into the picture as well. As a practical matter, we tend to care more about the number of tapes used than the total number of gigabytes. Dividing the backup into separate sets by server, application or other criteria makes it easier to manage your backups, but it consumes more tapes -- sometimes a lot more -- because you can't expect that each subset will match your tape capacity and you're probably going to have a lot of partially full tapes.

The big bugaboo is data growth. This is notoriously hard to estimate. Most vendors provide tools to help you estimate the size of your backup set to help you estimate your rate of data growth.

While the kind of tape technology you choose can make a difference, increasingly this is not a deciding factor in choosing a tape library. Many libraries will support several different drive types and companies like Quantum Corp. and Spectra Logic even offer libraries that will handle multiple drive types at the same time, partitioned as sub-libraries.

One factor that might be important in tape technology is the library's ability to be upgraded with new kinds of drives as they become available. A lot of libraries provide this feature if you're going to a new generation of the same technology -- say to LTO-5 from LTO-4 -- while some, such as a couple of models from Quantum, can be upgraded to use tapes with different form factors, such as DLT to LTO.


Most tape libraries are purchased through resellers. A good reseller can be an enormous help in selecting the right library. A not-so-good reseller can be a real problem. "Look at a reseller's depth of product line," Grover suggests. "Are you confident your reseller is giving you all the best choices? How much do you depend on reseller?"

Getting it right

Getting the right tape library isn't black magic, but it does take some care and thought. An expandable, modular tape library supported by sophisticated management features and tied to modern backup software will support your enterprise for years to come. It's worth taking the time and effort to make the right choice.

 About the author: Rick Cook specializes in writing about issues related to storage and storage management. 

Dig Deeper on Tape backup and tape libraries